Similarity measures are mechanisms that assign a numeric score indicating how closely two documents, or a document and a query match. The Cosine measure is one of the similarity measures that treat a document or a query as a vector of weighted terms or keywords. The similarity distance calculated by the Cosine measure is based on the exact matching of keywords. Thus the semantic relatedness between the keywords of the two documents is not considered.;The CSA framework improves the information sharing among agents that are semantically close to each other but whose vectors of key-phrases are syntactically different. The evaluation of CSA within the ACORN system demonstrates that the number of agents with which each agent shares its information increases compared to the original ACORN system. In the evaluation part, cliques of agents are used as criteria to show the performance of the new ACORN system after applying CSA. The agents that are very similar to each other join the same clique. The more populated the cliques are, the better the information sharing among the agents is. The average number of agents inside cliques increases after applying CSA to the ACORN system.;This thesis presents a category-based similarity algorithm (CSA) to determine the semantic similarity between any two pieces of information. The CSA is implemented inside the ACORN system and adds the semantic similarity feature to ACORN. CSA is applicable inside any information sharing system in which the pieces of information are represented as vectors of weighted keywords.
展开▼
